FR2812503A1 - Coding scheme for audience sampling uses spread spectrum audio data signal does not affect users - Google Patents

Abstract

A coding scheme inserts (B) a spectrum spread (A) digitized audio signal (EM) from a microphone with fixed length frames into an audio frequency sub band carrier for the sound signal using phase modulation and repeats this (D) to give a repetitive digital signal in the transmitted sound signal.

Description

l

  INFORMATION ENCODING AND DECODING METHOD AND SYSTEM

DIGITAL IN A SOUND SIGNAL

TRANSMITTED BY A REVERBERANT CHANNEL

  A method and system for coding-decoding digital information in a signal

  sound, transmitted by a reverberant channel.

  The management of the transmission of sound signals transmitted on audio and / or video frequency channels, currently involves the transmission of digital data, extrinsic to the audio signal or to the audio part of a video signal such as

television signal.

  This is in particular the case when it is necessary to provide users, listeners or viewers with information allowing either to improve the quality of the radio or television broadcasting service, or to promote pseudo-interactive game services with regard to questions / answers

predetermined.

  The solutions proposed for this purpose consist above all in detecting in the audiofrequency signal to transmit AS specific frequency bands in which this signal has a reduced level, in any case masked by the corresponding level of neighboring frequency bands, and to insert after filtering, by means of a band eliminator filter of the audiofrequency signal in the corresponding band or bands, a digital signal modulated in phase for example in the corresponding band or bands, in order to generate a digital signal

marked, marked AS *.

  Of course, due to the random nature of the amplitude level of the audio frequency signal in the aforementioned frequency bands, the elimination of this signal, and, consequently, the insertion of the digital signal in the latter, can only be achieved aperiodically. For this reason, the transmission of data in the form of digital messages can only be carried out asynchronously, in the form of digital frames. Such digital data transmission methods and systems have been described, in particular, in French patent applications No. 2,713,852, published on 06/16/1995 and No. 2,734,977 published on 07/25/1997 in the name of

TELEVISION FROM FRANCE.

  The methods and systems described in the aforementioned documents are satisfactory, in particular in the context of the aforementioned pseudointeractivity applications, in which a system for rendering and transmitting digital data is disposed and permanently placed with respect to a target receiver such as a television receiver, the system for reproducing and transmitting data being provided with an infrared and / or HF type link to a user console. The operating console then includes

  functions allowing the implementation of the pseudo-

interactivity, for the user.

  The data playback and transmission system can be coupled either by direct connection to a video recorder or a television receiver, if necessary a tape recorder, on which the audio frequency signal has been recorded, or by acoustic coupling by arrangement of the microphone. this system for restitution and transmission of data in the immediate vicinity of the loudspeaker of the television receiver or radio receiver, in order to recover the digital data transmitted from the sound signal, originating from the branded digital audio signal. In such acoustic coupling conditions, the aforementioned methods and systems have proved satisfactory, the system for restitution and transmission of data playing however the role of a transponder making it possible to ensure the transmission of digital data to the user console, whatever the position of the latter with respect to the television receiver or the radio receiver within a reasonable perimeter, in which the sound signal transmission channel does not constitute a reverberant channel. The subject of the present invention is the implementation of a method and a system for coding-decoding digital information in a sound signal transmitted by a reverberant channel, in which any implementation of an intermediate restitution system and transmission of digital data between a target receiver, television receiver or radio receiver, and the

  device or user console is deleted.

  Another object of the present invention is, in

  Consequently, the implementation of a coding process-

  decoding of digital information in a sound signal allowing the execution of the decoding process in a nomadic user device or console, the transmission of digital data to this user device or console being carried out from the target receiver via the single signal sound in a reverberant environment. Another object of the present invention is also the implementation of a coding-decoding system of digital information in a sound signal in which the user device, made completely independent of the target receiver, allows the implementation of the method. decoding according to the invention in the absence of any constraint vis-à-vis other users of this

target receiver, or without their knowledge.

  The method of coding-decoding digital information in a sound signal, object of the present invention finds application to the provision of management or operating services of radio or television networks of data transmission networks, in particular for the implementation of audimetry services, the totally independent character of the nomadic user device being able, in a particularly advantageous manner, to be used to ensure functions of audience sampling or of monitoring with respect to target receivers, such as radio and / or television receivers not dedicated to such functions, in the absence of any imposed constraint

to users of these.

  The method and the system for coding digital information in a sound signal, objects of the present

  invention, this sound signal being generated by the loudspeaker

  speaker of a target receiver, excited by an audio frequency signal, and received by a receiving microphone, are remarkable in that it consists in, respectively allows, to establish this digital information in the form of frames of fixed length, each frame comprising at least one useful data field, the frames constituting a digital message being subjected to a spectrum spread processing on a determined number of bits to generate a spread spectrum digital message, insert the

  spread spectrum digital message in at least one sub

  audio signal frequency band, by phase modulation around a carrier inserted in this frequency sub-band from the spread spectrum digital message, periodically repeat the insertion of the spread spectrum digital message for at least one frame for generate a repetitive digital message inserted in the sound signal. By spread spectrum digital message, we actually mean the signal

  digital spread spectrum message carrier.

  The method and the system for decoding spread spectrum digital messages, coded and inserted into a sound signal, in accordance with the method which is the subject of the present invention, consists in, respectively enabling, upon reception of at least one spread spectrum message consisting by a digital frame, each digital frame being subjected to demodulation and decoding from a spread spectrum sequence, calculate for at least one of the data fields of each digital frame the value of the function d autocorrelation between the demodulated message on this field and the spreading sequence used on transmission, compare this value of the autocorrelation function with a determined threshold value, use as value of the data field code the value of this field if the value of the autocorrelation function is greater than this threshold value,

  otherwise generate a false detection signal.

  The method and system for coding-decoding digital information in a sound signal, objects of the present invention, will be described in more detail below, in conjunction with the figures below in which: - Figure 1 shows a general flowchart, illustrating the steps allowing the implementation of the method of coding a digital signal into a sound signal, in accordance with the object of the invention; FIGS. 2a1, 2a2 and 2a3 represent, with reference to the flowchart represented in FIG. 1, a detail of implementation of spread-spectrum coding of a bit stream, the modulation of a carrier from 'a resulting sequence, respectively the constitution of digital frames with spread spectrum from the coded bit stream; - Figure 2b shows, with reference to the flowchart shown in Figure 1, a process of inserting a digital signal into the audio signal supporting the sound signal, this insertion process being carried out from a step of filtering of at least one frequency band, in order to eliminate this frequency band, then insertion into the eliminated frequency band of a carrier wave modulated in phase around the central frequency of this eliminated frequency band; - Figure 3a shows a general flowchart illustrating the steps for implementing the method of decoding a digital signal inserted in a sound signal, in accordance with the object of the present invention; FIG. 3a2 represents, by way of illustration, a process for decoding useful digital data contained in a useful data field of a digital frame with spread spectrum fixed length received; - Figure 3b shows an illustrative flowchart of a preferred embodiment of a decoding method as illustrated in Figure 3aj, in which a majority vote type error correction process is introduced; - Figure 4al shows, by way of illustration, a functional diagram of a system for coding digital information in a sound signal, in accordance with the object of the present invention; - Figure 4a2 shows a detail of non-limiting embodiment of a spread spectrum processing module and an insertion module by carrier modulation allowing, by switching, the constitution of digital messages spread spectrum formed by a frame of fixed length comprising a synchronization word and at least one useful data field; - Figure 4a3 shows an equivalent functional diagram of the spread spectrum processing module and the insertion module shown in Figure 4a2, when the latter, switched to a first position, position I, ensure, for the constitution of each frame of fixed length, the introduction of a synchronization word; - Figure 4a4 shows an equivalent block diagram of the spread spectrum processing module and the insertion module shown in Figure 4a2, when the latter, switched to a second position, position II, ensure, for the constitution of each frame of fixed length, a spread spectrum treatment of one or more useful data fields; FIG. 5a represents, by way of illustration, a functional diagram of a system for decoding digital information in a sound signal, in accordance with the object of the present invention, this decoding system allowing, on the one hand, ensure detection of the reception of each frame of fixed length, from the synchronization word, then, furthermore, following this detection, a decoding of the useful data contained in the useful data field or fields of the frame of fixed length considered; FIG. 5a2 represents an equivalent functional diagram of the decoding system, object of the present invention, represented in FIG. 5al, in the more specific case where the latter is configured, by switching, to ensure the detection of the synchronization word included in each fixed length spread spectrum digital frame; - Figures 5a3 to 5a8 show different timing diagrams of autocorrelation function values of the synchronization word, in an optimized calculation process making it possible to minimize the rate of false detections of the synchronization word; - Figure 5b shows an equivalent functional diagram of the decoding system object of the present invention shown in Figure 5al, in the more specific case o, following the detection of the synchronization word of a frame, the latter is configured, by

  switching, to ensure the decoding of the useful data.

  A more detailed description of the

  coding, respectively decoding of digital data in a sound signal transmitted via a reverberant channel between the loudspeaker of a target receiver, such as a radio receiver or a television receiver, and the microphone of a device nomad extraction of this digital signal, will now be given in connection with Figure 1 and the following figures.

  Prior to the actual description of

  coding method, respectively decoding, object of the present invention, a reminder of the definition and definition parameters of a transmission channel of

  reverberant type will be given below.

  Generally, it is indicated that the transmission channel of a sound signal includes all the elements of the transmission chain placed between the source, the speaker, and the recipient, i.e. the microphone. , the aforementioned transmission channel therefore comprising the aforementioned loudspeaker, ambient air and microphone. In general, a representation of the aforementioned transmission channel is given by the expression of the signal received by the microphone according to the relation (1): r (t) = s (t) h (t) + b (t) + be (t) In this relation: r (t) denotes the aforementioned received signal, s (t) denotes the sound signal emitted by the loudspeaker, h (t) denotes the channel transfer function, b (t) denotes white noise which is superimposed on the sound signal itself during transmission by the latter by the acoustic channel to thus constitute the received signal, be (t) denotes the contribution of external acoustic noise, corresponding to ambient noise, 0 denotes the contribution of the channel transfer function to the transmission of the signal s (t) produced by

convolution of s (t) and h (t) ,.

  In general, it is indicated that the acoustic channel is characterized by a multipath propagation linked to the multiple reflections of the incident sound wave on the reflecting surfaces of the environment in which the speaker and the microphone are located. It therefore constitutes a so-called channel

reverberant.

  Due to the aforementioned reverberation phenomenon, the sound waves interfere and, depending on their phase relationship, attenuate or amplify each other. Due to the aperiodic change in these phase relationships, depending on the location and the configuration of the latter, interference phenomena occur and a wave phenomenon

  most often can be highlighted.

  In particular, when using a device for extracting the digital signal of nomadic type, the movement of this device causes the latter to receive waves of variable amplitude and phase, the reverberant channel then behaving as

  a channel fluctuating with time.

  In general, the transmission channel can be characterized by its impulse response, which depends on time, during the movement of the extraction device or even of the activity and the

  configuration of the ambient environment for example.

  The Fourier transform of the impulse response according to a variable T for spreading the delays of the different paths represents the transfer function of the channel as a function of the frequency. This transfer function, designated by R.T.F for Room Transfer Function, depends on the characteristics of the environment and the room in which the acoustic wave is propagated, as well as on the place in the room o

  the aforementioned sound signal is received.

  The speed of variation of the characteristics of the transmission channel can then be characterized by its autocorrelation function. This function makes it possible to identify the parameters of the channel such as the coherence band of the channel, denoted Bc, or the coherence time of the

channel, noted Tcc.

  The coherence band Bc can be defined as the frequency bandwidth such that a signal with a bandwidth significantly less than this coherence band has all of its spectral components affected in the same way by the channel considered. The coherence band Bc is proportional to the inverse of the spreading of the delays T, which, in the application considered for the implementation of the method which is the subject of the present invention, can be at most of the order of a second . Under these conditions, the coherence band Bc is

substantially equal to 1 Hz.

  Thus, a sound signal whose frequency band is greater than the coherence band of the channel transmitting the latter is the site of distortions and interference between symbols in the case of digital communication. We then speak of selectivity

  frequency of the transmission channel.

  Frequency selective channel In the application considered by the object of the present invention, the transmission channel will not be considered as frequency selective if the bandwidth B of the sound signal is clearly less than the coherence band Bc. Assuming that the bit rate D, expressed in bauds, of the transmission channel corresponds approximately to the bandwidth of the sound signal B expressed in Hz, the upper limit of the authorized bit rate is given by:

D <1 baud.

  Thus, for applications in which the gross transmission rate must be significantly greater than the minimum rate indicated above, the reverberating transmission channel constituting in this case a frequency selective channel, the method which is the subject of the present invention must

  necessarily take into account the aforementioned constraint.

  Time-selective channel The multiple dispersions and reflections in the immediate vicinity of the receiver of a sound signal into which digital data have been inserted generates, in the vicinity of the latter, a multitude of elementary waves. For each propagation path, the signal arriving at the receiving microphone, that is to say the device for extracting the digital signal, is composed of a whole series of signals whose propagation time differences remain small compared to the duration of symbol time but significant in relation to the period

of the carrier.

  During a displacement of the aforementioned receiver, the random variations in amplitude and phase of the different echoes result in selective fading over time. These fading results in random fluctuations of the signal with dropouts spaced about a half at a wavelength, that is to say according to the speed of movement of the receiver every Toc seconds. These time-selective fading effects are the cause of synchronization losses, large untimely drops in the signal-to-noise ratio, which of course contributes to the introduction of bit error rates.

  high unacceptable at reception.

  Consequently, the coherence time TcC of the transmission channel can be defined as the duration during

  which characteristics of the latter vary little.

  This coherence time is proportional to the inverse of the Doppler frequency band fd according to the rate of

  correlation chosen to characterize the aforementioned channel.

  In the application relating to the implementation of the coding-decoding method which is the subject of the present invention, it is indicated that the Doppler frequency can take as maximum value fd max such that: fd max z 40 Hz. The above-mentioned maximum Doppler frequency is obtained for a speed of movement of the receiving microphone, or of the nomadic extraction device, substantially equal to 1.4 m / s, a wavelength of the sound signal given by the speed of propagation of the acoustic wave in the air substantially equal to 340 m / s for a maximum frequency of the sound signal taken equal to 10 kHz, the movement of the receiving microphone and the portable extraction device, transported by the user for example, being considered to be collinear with the angle affect

  of the sound wave transmitted to the latter.

  Under these conditions, the minimum coherence time

  Tcc of the transmission channel is approximately Tcc 0 12.5 ms.

  The transmission channel is said to be time selective when the signal analysis time Ts

  sound is greater than the coherence time Tcc.

  The transmission channel is said to be temporally non-selective if, on the contrary, the duration Ts is much less than the coherence time Tcc. In the latter case, it is considered that the attenuation and the phase shift brought by the transmission channel to the propagation of the sound signal remain constant during the analysis time.

Ts of the information symbol.

  In view of the above and for

  the application to the implementation of the coding process-

  decoding of a digital signal inserted in a sound signal, object of the present invention, it is indicated that the channel will be considered as non-selective in time if the information symbol duration remains less than:

Ts <Tcc or Ts <125ps.

  Under these conditions, the minimum value Ds of the transmission rate through the channel must be greater than:

From 2 8 kbaud.

  With regard to the noise affecting the transmitted sound signal, it is indicated that the thermal noise coming from the components of the system could be neglected due to the fact that the digital signal must be recovered

  when the audible signal remains audible.

  However, with regard to external noise or ambient noise which is superimposed on the sound signal, since it comes from different sources of noise in the environment in which the receiving microphone is located, this noise component will be significant in the case of narrow band noise compared to the

  bandwidth of the audio signal.

  Consequently, the constraints relating to the reverberant transmission channel authorizing the implementation of the coding-decoding method object of the present invention

can be summarized below.

  The choice of the characteristics of the data signal, and therefore of the digital signal inserted in the sound signal, such as the bit rate of the latter, must be made according to the techniques to be used to adapt the mobile device extracting the inserted digital signal

  in the audio signal, to the transmission channel.

  Under these conditions, so that the channel is not frequency selective, the bandwidth of the data signal B must be much less than the coherence band Bc, which imposes a maximum bit rate Ds defined by:

Ds <1 baud.

  Furthermore, for the transmission channel to be non-selective in time, the duration of an information symbol time must be much less than the channel coherence time Tcc, which imposes a minimum bit rate defined by:

From 2 8 kbaud.

  In the context of the targeted applications, the symbol rate Ds must therefore be greater than 1 baud and the transmission channel can therefore be considered to be frequency selective. Furthermore, if the symbol rate Ds is chosen to be less than 8 kbaud, the transmission channel may

  also be considered time-selective.

  Under the aforementioned conditions, the propagation of the sound signal into which the digital data is inserted is faced with the worst case that can be envisaged during a communication.

digital.

  The subject of the present invention is the implementation of a coding-decoding method of a digital signal in a sound signal making it possible to obtain satisfactory transmission and extraction of this digital signal, and in particular to information conveyed by the latter when using

nomadic extractor devices.

  The method of coding digital information in a sound signal, object of the present invention,

  will now be described in conjunction with FIG. 1.

  In general, it is recalled that the sound signal is generated by the loudspeaker of a target receiver, such as a radio frequency receiver, a television set, a multimedia computer or the like, excited by an audio frequency signal marked and received by a reception microphone fitted with a device for extracting digital data, the sound signal thus being transmitted via a reverberant channel. Thus, we have the audio signal, denoted AS, digital audio signal representative of the sound signal, into which digital data must be inserted, to generate a signal

digital audio marked AS *.

  With reference to the same FIG. 1, the coding method which is the subject of the invention consists, in a step A, of establishing the digital information in the form of frames of fixed length, each frame, denoted Tr, comprising at least one field of useful data. The frames constitute a digital message, denoted DM, consisting for example of two or more words of useful data di, d2. From this digital message, a digital signal is generated subjected to a direct sequence spread spectrum treatment from a spread spectrum sequence denoted ESQ. The spread spectrum processing operation is noted:

EM = DM 0 ESQ

  o EM denotes the resulting sequence obtained by the processing operation, o denotes an amplification and amplification operation in phase of the digital message DM and of the spreading sequence ESQ, as will be described in more detail in the

description.

  In addition to the above-mentioned spread spectrum processing operation, the digital message DM can be subjected to M-ary coding, by choosing the spectrum spread sequence chosen, from among a plurality of M spectrum spread sequences. determined, as will be described in more detail later

in the description.

  The aforementioned step A is followed by a step B consisting in inserting the resulting sequence EM into at least one sub-frequency band of central frequency fi of the audio signal AS, by phase modulation

  around a carrier, noted CSBi, inserted in the sub-

  aforementioned frequency band. The phase modulation is carried out from the resulting EM sequence under the conditions which will be explained later in the

  description and allows to generate a modulated carrier,

  noted nCSBi. The operation of inserting the resulting sequence EM into one of the frequency sub-bands of the audiofrequency signal AS is symbolically noted in step B by the relation: AS * = AS + cCSBi o AS denotes the audiofrequency signal in which the

  center frequency sub-band fi has been deleted.

  Indeed, the above-mentioned insertion operation in step B can advantageously consist in filtering the audio frequency signal AS to eliminate a frequency band of central frequency fi and of determined bandwidth, this filtering operation, carried out from a band eliminator filter for example, which can then be followed by an actual insertion step of the nCSBi modulated carrier. The aforementioned insertion step can consist of adding the modulated carrier frequency signal tCSBi to the digital audio signal.

previously filtered, AS-.

  Step B is then followed by a step C consisting in transmitting the digital audio signal marked AS *, in which the digital information has been

  inserted through the resulting EM sequence.

  The transmission carried out in step C can be carried out, for example, by frequency modulation of a conventional type radio wave, and, for this reason,

will not be described in detail.

  Of course, and according to an advantageous aspect of the method of coding digital information in a sound signal, object of the present invention, the process of inserting the resulting sequence EM is repeated periodically for at least one frame Tr to generate a message with repetitive spread spectrum inserted in the audio signal AS and in the sound signal. The repetition operation is denoted D by a return arrow from the end of the insertion step to step B at the start of the same step B. It is understood in particular that the periodic repetition for the same frame, the if necessary for a set of successive frames, makes it possible to generate a repetitive message inserted in the audio frequency signal

  marked and in the sound signal thus transmitted.

  With regard to the implementation of steps A, B, C and D shown in FIG. 1, it is indicated that the sub-steps relating to the establishment of the digital information in the form of frames, to the processing by spread spectrum , if necessary to M-ary coding, can be carried out prior to step B, the

  frames thus constituted in digital form, that is to say

  say by successive binary values, which can be memorized, the sub-steps of filtering at the central frequency fi and of modulation around this central frequency of a carrier CSBi at the same central frequency fi which can then be carried out successively. The phase modulation step of the carrier CSBi around the central frequency fi can then be carried out from

  the resulting sequence memorized in step A above.

  A more detailed description of a process

  specific for performing spectrum spreading processing and M-ary coding of the digital message DM in order to constitute the digital frames Tr previously mentioned for the implementation of step A of FIG. 1, will now be given in connection with the

Figures 2a1, 2a2 and 2a3.

  In FIG. 2aj, on the one hand, a digital message DM has been represented in its temporal form x (t) as

  a succession of bits at the value 1 respectively -1.

  In the representation of FIG. 2a1, the digital message DM constitutes a bit stream of bit duration Td. In accordance with a remarkable aspect of implementation of the coding method, object of the present invention, the digital message DM is then subjected to a spectrum spreading process from a spreading sequence, denoted ESQ, represented at Figure 2a1 in its time form g (t). Remember that this type of

  signal generally consists of a pseudo- sequence

  bit random at the value 1 or -1, also called burst or chip, of duration Tc. The pseudo-random spectrum spreading sequence corresponds to binary elements

  of value +1 or -1, of chip duration, elementary duration Tc.

  Taking into account the aforementioned spectrum spreading process, for a transmission of a signal modulated by direct sequence spreading spectrum of the form g (t), the insertion and the transmission of the signal after spreading spectrum process can be performed by a BPSK type phase modulation, the sound signal is received with a delay echo T via the reverberant channel, a narrowband type I (t) jammer being superimposed via this same channel reverberant, the sound signal received verifies relation (2): In this relation, it is indicated that: fp = f represents the frequency of the modulated carrier transmitted in step C, a 'represents the attenuation of the echo signal by relative to the direct path, is a random phase uniformly distributed over

the interval [0.2n].

  The operation of despreading the spectrum of the received signal r (t) then consists in multiplying the reception signal by the spreading sequence g (t) used on transmission. Provided that the spreading sequence reproduced is perfectly synchronized with that contained in the sound signal received on the direct path, then the signal obtained after despreading checks the relation (3):

  m (t) = X (t) g2 (t) COS (27cfpt) + ax (t-T) g (t) g (t-T) cos (21fpt + 0) + g (t) I (t).

  In this relation, we indicate that g2 (t) = 1.

  If, moreover, the delay T introduced by the multiple reflections is greater than the duration of the chip Tc, then, the relation (4) verified by the pseudo-random sequence delayed by the delay t with respect to the random sequence transmitted by the direct path is written: Relation 4 g * (t) g (tT) dt "g * (t) g (t) dt and shows that the energy of the spread spectrum sound signal transmitted by multiple reflections is much lower than the energy of the spread spectrum sound signal transmitted by

the direct route.

  It can thus be seen that the spectrum spreading process, introduced in accordance with the method which is the subject of the present invention, makes it possible to eliminate the linked interference.

multipath.

  Furthermore, the noise signal I (t) being spread over a much wider frequency band than that of the audio signal x (t), during the reception filtering and demodulation operation, an important part of the The energy of the noise signal is thus eliminated. Thus, the spread spectrum processing process implemented in step A illustrated in FIG. 1 previously cited, according to the methods indicated in relation to FIG. 2aj, is perfectly justified and

  eliminates multi-interference phenomena

trips to reception.

  Of course, the modulation by the constituent chips of the pseudo-random sequences as represented in FIG. 2al makes it possible to obtain a digital message with spread spectrum, combination of the bit values

  of either of the two signals.

  The resulting spread spectrum sequence then checks the relationship:

EM = DM 0 ESQ.

  FIG. 2a2 represents the modulation operation of the carrier CSBi not modulated by the above-mentioned resulting sequence EM. The modulation operation means a PSK type phase modulation operation, the phase of the unmodulated carrier CSBi being phase shifted by 0 or t as a function of the value of +1 respectively -1 of the resulting sequence EM to get the modulated carrier

iCSBi shown in Figure 2a2.

  Obtaining the carrier modulated by the resulting sequence allows the transmission of spread spectrum digital frames, the aforementioned frames thus being able to be formed in accordance with the coding method object of

  the invention, as shown in Figure 2a3.

  In general, it is indicated that each frame Tr can comprise at least, advantageously, a word

  synchronization and at least one useful data field.

  In the context of this description,

  indicates that the term data field of each frame indifferently designates either the synchronization word as such, or the field relating to the useful data transmitted, that is to say to the data allowing the management of the services offered by

through the audio signal.

  Thus, with reference to FIG. 2a3, the synchronization word MS constitutes the data field Do and the useful data word the data fields Dl and D2, for

  a digital spread spectrum frame considered.

  For the constitution of the aforementioned fields, with reference to FIG. 2al, it is indicated that the technique of spread spectrum by direct sequence can be used, the length of the sequence in number of chips

being chosen equal to Lc = 31.

  Digital frames and digital spread spectrum messages are thus transmitted with a chip rate Dc = 175 chips / s, the width of the occupied band being

  under these conditions of approximately B = 350 Hz.

  In addition, M-ary coding of the payload can be applied to increase the transmission rate of the spread spectrum digital message ultimately transmitted. For this purpose, each symbol, and in particular on each channel when the transmission is carried out on a phase channel and on a phase quadrature channel, as well as

  will be described later in the description, can thus

  be spread by a sequence chosen from a number M = 8, for example of Gold sequences, coding four bits

  useful in keeping the sign bit.

  It is thus understood that the step of spectrum spreading processing and of M-ary coding can advantageously consist in applying to the carrier CSBi, during the modulation of the latter by the resulting sequence EM, a spectrum spreading sequence chosen from direct sequence spread spectrum sequences. According to a remarkable aspect of the coding method which is the subject of the invention, the type of spreading sequences can be chosen as a function of the field of each frame constituting the aforementioned digital signal. Thus, the spectrum spreading sequences can be chosen from the group of pseudo-random sequences of the Gold sequences or of the Walsch-Hadamard sequences, or others. In an embodiment corresponding to the choice of a sequence from a number of eight Gold sequences coding four useful bits, previously mentioned, each frame advantageously had a start or synchronization word Do of length Lc = 31. This start word was followed by two words D1, D2 of coded payload

  by Gold sequences of length Lc = 31 chips.

  The use of 8 Gold sequences makes it possible to code 4 bits, 3 bits being linked to the code and one bit linked to the sign, per channel, that is to say 8 bits useful when the modulation around the central frequency fi for the insertion of the modulated carrier nCSBi is a phase modulation MDP4

  consistent for each data word.

  The M-ary coding implemented makes it possible to increase the bit rate to be transmitted. This type of coding makes it possible to code the useful information bits by using several spreading sequences. To code M useful bits, 2M

sequences are required.

  In an exemplary embodiment and for an MDP4 modulation on a channel in phase I and on a channel in quadrature of phase Q, on each channel it is decided to code 3 bits by 8 different Gold codes. A fourth bit is further coded by signing with + or - the sequence

spreading.

  The following coding rule can be used for a series of binary elements below:

  b0, bl, b2, b3, b4, b5, b6, b7, b8, b, bl0, bl, bl2, bl3, bl4, bl.

  The original frame being made up of two data words d1, d2, one data word makes it possible to code 8 useful bits. The three even bits are coded by a spreading sequence among 8 spread spectrum sequences denoted codeki (t) with k E [0 ... 7], the fourth even bit codes the sign of the sequence used for the channel in phase I: b0o, b2, b4 - choice of a code among 8 for the channel in phase I; b6 -> code sign for the channel in phase I. The three odd bits are coded by a spreading sequence among 8 spread spectrum sequences, denoted code_k_q (t) with ke [0 ... 7], the fourth odd bit codes the sign of the sequence used for the phase quadrature channel Q: bl, b3, b5 - choice of a code from 8 for the channel in phase I; b7 -> code sign for the quadrature phase Q channel. The 8 sequences used for channel I and channel Q can advantageously be different, which implies the use of 16 different spectrum spreading sequences. The rule for coding the sequences as a function of the useful bits is chosen in the application according to the table Tl below:

TABLE T1

  Index of the code k used b0 or b. b2 or b3 b4 or b5 Code k i (t) Code k q (t)

-1 -1 -1 0

-1 -1 +1 1

-1 +1 +1 2

-1 +1 -1 3

+1 +1 -1 4

+1 +1 +1 5

+1 - +1 6

+1 -1 -1 7

  For example, to code the following series of binary elements:

  -1, 1, -1, 1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1

  ò the first data word uses as spreading sequence: - on the phase path: (+1) x codeli (t) which codes the bits bo = (- 1), b2 = (- 1), b4 = l with the sign (+1) coding the bit b6 = 1; - on the quadrature channel: (-1) x code_4_q (t) which codes the bits b1 = l, b3 = l, b5 = (- 1) with the sign (-1) coding the bit b7 = (- 1) ; * the second data word uses as spreading sequence: - on the phase path: (+1) x code_5_i (t) which codes the bits b8 = l, blo = l, b12 = l with the sign (+1 ) coding the bit b14 = l; - on the quadrature channel: (-1) x codeo0_q (t) which codes the bits b9 = (- 1), bll = (- l), b13 = (- l) with the sign (-1) coding the bit

b15 = (- l).

  In the previous relationships, the symbol x represents

the multiplication operation.

  Thus, with reference to FIG. 2a3, each frame

  Tr therefore carries a total of 16 useful bits.

  A more detailed description of the operations

  necessary for the implementation of step B of insertion of the resulting sequence EM into the audio frequency signal AS

  will now be given in connection with FIG. 2b.

  The aforementioned step B in fact consists in inserting the frames, as described in FIG. 2a3, in at least one of two low frequency sub-bands whose central frequency is between 2.5 kHz and 4 kHz and in at least one of four high frequency sub-bands between 4 kHz and the cut-off frequency of the

  audio signal and sound signal.

  With reference to FIG. 2b above, for any frequency band of central frequency fi whose energy level in dB is less than an AM masking value, this AM masking value possibly being relative to the value of the energy of lateral bands driving the central frequency band fi as shown in FIG. 2b, the filtering operation by a band eliminator filter causes a "hole" in which

  the signal is suppressed around the center frequency fi.

  The insertion of a carrier of the same central frequency fi but phase-modulated, denoted tCSBi, and of a lower energy level, of course, than the AM masking value, makes it possible to insert the digital message with spread spectrum corresponding to each frame in the audio signal

  to generate the audio signal marked AS *.

  With regard to the conditions necessary for implementing the operating process relating to the insertion of the digital spread spectrum message into the audio frequency signal AS, reference may usefully be made to French patent applications No. 2 734 977 and

  No. 2,713,852 mentioned previously in the description.

  Indeed, the conditions relating to the masking values and the bandwidths as well as to certain values of central frequency fi can be included at least in part. However, as regards the choice and the number of sub-bands used, the coding method which is the subject of the present invention has been able to be implemented.

  advantageously from six sub-bands.

  From a practical point of view, it has appeared advantageous to transmit data frames in three sub-bands simultaneously, one sub-band among the low frequencies and two sub-bands among the high frequencies. By way of nonlimiting example, the values of central frequency fi are designated in table T2 below, o i denotes the rank of the subband and fc the value of the central frequency fc corresponding to the

  value fi of FIG. 2b for example.

TABLE T2

LOW FREQUENCIES HIGH FREQUENCIES

  i = 15 or f0 = 2925 Hz i = 18 or f0 = 4850 Hz i = 16 or fc = 3425 Hz i = 19 or f0 = 5850 Hz i = 20 or f0 = 7050 Hz i = 21 or fc = 8600 Hz In this which relates to the bandwidth of the

  modulated carrier nCSBi inserted in each of the

  frequency bands fi, the selected width was B = 350 Hz around each central carrier frequency fi-. In general, it is recalled that the masking models used can use the psycho-acoustic model described in the patent applications

  French previously mentioned in the description. This

  model is inspired by the MPEG specification and the modifications introduced relate only to an adaptation of the calculation of the mask and of the cutoff filter of the band eliminator type of width B = 350 Hz adapted to the bit rate of 175 chips / s. Listening tests on the analyzed sound sequences showed inaudibility

  of the coding method which is the subject of the present invention.

  Analogously to a specific choice of coding by spread spectrum by direct sequence applied as a function of the nature of the fields of the frames considered, the type of phase modulation applied to the carrier CSBi, then representative of the resulting sequence EM to be inserted, is also a function of the data field considered, that is to say the field Do relating to the synchronization word, respectively the fields Dl and D2

relating to useful data.

  Under these conditions and as illustrated in FIG. 2a3, the chosen phase modulation applied to the carrier CSBi is a phase modulation MDP2 for the synchronization word MS with reference to FIG. 2a2 and a phase modulation MDP4 for the useful data of the

fields Dl and D2.

  Both with regard to the phase modulation carried out in step B around the central frequency fi and at the reception to ensure decoding, as well as

  will be described later in the description, the follow-up

  of the phase can be achieved by a Costas loop of the MDP2 type for the synchronization word MS and of the MDP4 type for the useful data words D1 and D2. The Costas loop implemented can be a first order loop. The removal of the phase ambiguity of value n during the implementation of the phase modulation MDP2 for the synchronization word MS can be carried out from the sign of the correlation peak noted on the start word, as well as it will be described later in the

  description, relative to the decoding process object of

* the present invention.

  A more detailed description of the

  decoding of digital information in a sound signal, object of the present invention, will now be given in

  connection with FIG. 3al and the following figures.

  With reference to FIG. 3 aforementioned, and due to the repetition of N successive frames [Trk] N identical, there is available at the output of the speaker of the target receiver the aforementioned sound signal in which the frames or digital messages with spread spectrum previously mentioned have been inserted successively. Thus, for a given message, there are available at reception E of N identical frames in order to allow the implementation of the decoding method which is the subject of the present invention. The successive frames received are subject to decoding and demodulation, as shown in FIG. 3a2. For this purpose, the carrier nCSBi modulated by the resulting sequence EM, either by the digital message and by the spreading sequence ESQ, is subjected to a decoding denoted nCSBi ESQ by the spreading sequence ESQ. We obtain a carrier modulated by the only original digital message do, dl and d2. A demodulation of the latter makes it possible to restore the digital message, ie the fields

do, dl and d2.

  With reference to FIG. 3al, this consists in calculating in step F, for at least one of the data fields of each digital frame Trk, the value of the autocorrelation function between corresponding fields of the successive frames. The value of the autocorrelation function is denoted ACF (Di). In step F, the correlation calculation is carried out for all the sub-

  used carriers of central frequency fi.

  The operating mode of the decoding method which is the subject of the present invention thus makes it possible to demonstrate an operating mode using the synchronization word MS positioned at the start of the frame, field Do, constituted by a single spreading sequence having good properties of autocorrelation, in order to make it possible to perform on the following symbols, that is to say on the symbols represented by the useful data fields D1 and D2, a substantially synchronous detection greatly optimizing the correlation margin and, consequently, the

  performance of the whole system.

  Furthermore, once the synchronization word MS has been detected upon reception, levels of the correlation peaks on the following symbols can be measured and compared with threshold values below which the procedure for detecting false detections previously

  mentioned can be implemented.

  An adjustment of the threshold values makes it possible to have a more or less low probability of false detection of frames. As shown in FIG. 3a1, a step I of detection of the synchronization word MS comprising the step E of reception of the N identical successive frames and of demodulation and decoding, followed by step F of calculation of the autocorrelation function on the synchronization word MS, that is to say the data field Do, delivering the value of the autocorrelation function ACF (D0) may consist, in a step G, in comparing the above-mentioned value ACF (Do), by comparison of superiority, at a synchronization threshold value S0. This threshold value can be between 0.20 and 0.26, i.e. 0.20 <S0 <0.26,

  the comparison being noted ACF (Do)> So.

  On a negative response to the comparison test G, no frame having been detected in the insertion sub-bands considered, a step H makes it possible to search for another possible frame by return HE to the step of

  reception E by time shift of the input signal.

  On a positive response to the comparison test G, the presence of a frame having actually been detected, a step I is triggered in order to initiate the mode of detection of the data of the fields D1 and D2, the synchronization being acquired. Step I is followed by a step J for calculating correlation on the data word D1 and D2. This step is carried out in a similar manner to that of step F, the values of the autocorrelation function, calculated on the data fields Dl and D2 being, for this reason, denoted ACF (D1) and ACF (D2) respectively. . The ACF (Dl) and ACDF (D2) values used are the maximum values, as will be described later in the

description.

  Step J is followed by a step K consisting in comparing the aforementioned values of the autocorrelation function ACF (D1) and ACF (D2) with a threshold value called false detection. For a false detection threshold value valid for the data fields D1 and D2, and, for this reason, denoted S12, between 0.03 and 0.5, the comparison step K, denoted ACF (D1) and ACF (D2) <S12, a positive response to the inferiority comparison test K calls for a step L of reporting a false detection. On the contrary, a negative response to the inferiority comparison test K calls a step M of validation of the code value of the received signal, noted:

VD1 = D1 VD2 = D2.

  Taking into account these observations, it is indicated that the structure of the frames is adapted to ensure the best

detection possible.

  For this reason, in the example previously given, the synchronization word MS has a chip length L, = 31 corresponding to a privileged pseudo-random sequence PN. Such a sequence has good autocorrelation properties, typically N at rank 0 and -1 elsewhere, while on the other hand, it has low intercorrelation values, typically the value -1 with Gold sequences generated for this same sequence. Thus, the synchronization word MS, that is to say the field Do, is followed in the frame of the useful data fields D1 and D2, which are then, preferably from useful words of origin di, d2 , coded either by Gold sequences of chip length Lc = 31, or by

  Walsch-Hadamard sequences of chip length Lc = 16.

  The transmission mode carried out in step C of FIG. 1 with start or synchronization word MS makes it possible to code an additional bit depending on the sign of the transmitted code, and therefore on the sign of the correlation carried out on reception on the data word. . Under these conditions, and thanks to the introduction of M-ary coding, the use of four spread spectrum codes or sequences makes it possible to code three bits, two bits being linked to the code and one bit being linked to the sign, and the use of eight spread spectrum codes or sequences makes it possible to code four bits, three bits being linked to the code and one bit being linked to the sign. A code is a series of random bits constituting the spectrum spread sequences considered. The specific structure of the frames implemented to ensure the coding of the digital signal in the audio frequency signal and the sound signal in accordance with the coding method which is the subject of the present invention, makes it possible, in reception, that is to say for setting work of the decoding method, to highlight two successive operating modes: - the frame synchronization mode, o for example only the value of the code of the synchronization word MS is detected; - a mode for decoding useful data, o all of the code values capable of being used on transmission for coding useful data fields

Dl and D2 is then implemented.

  Thus, during the implementation of the decoding method which is the subject of the invention, for successive frames containing a specific synchronization word MS, the value of the threshold Sv constitutes a value of synchronization threshold. It can be between Sv = SO = 0.20 and 0.26 depending on the nature of the acoustic channel and the properties of the latter and whether or not an error correction process by mode voting is introduced majority. With regard to the threshold value applied to the detection of the useful data, that is to say to the threshold value Sv = S1 or Sv = S2 applied during the calculation of the corresponding autocorrelation function for the fields D1 and D2 of each frame, the corresponding threshold value constitutes a false detection threshold value. It is then between 0.03 and 0.05 depending on the sequences transmitted by the acoustic channel, in the absence of implementation of a correction process

  majority vote error.

  As regards the calculation of the values of the autocorrelation functions ACF (Di), the calculation implemented is a calculation of the conventional type by comparison of the values of the signal samples received with respect to the samples of the known coding sequences . The spread spectrum code or sequence giving a maximum correlation peak is retained during the relative displacement of the code contained in the frames and of the code of the known sequences for calculating the value of

the autocorrelation function.

  Of course, the advantage of repeating identical frames, as mentioned previously in the

  description, lies in the possibility of performing, for the

  implementation of the decoding method which is the subject of the present invention, an error correction process of the type to

majority vote.

  Such an operating mode is shown in FIG. 3b. According to a particularly advantageous aspect of the decoding method which is the subject of the present invention, within the framework of the implementation of the majority vote type error correction process, the determined threshold value constituting the synchronization threshold value S0 is preferably lowered to the value equal to or less than 0.20 so as to allow the decoding of a larger number of frames at the risk of

detect false frames.

  The decoding of the useful data of the fields D1 and D2 can then be carried out by correlation of N = 2P codes with the useful data, P designating the number of bits coded in the frame. The signed levels of the correlation peaks obtained by calculating the values of the autocorrelation functions of each code and of the useful data can be integrated on a number M of given words, the number M defining the depth of the vote.

majority mentioned above.

  In general, it is indicated that no process of normalization of the calculated values is carried out. More specifically, the detection of data words from fields D1 and D2 associated with a majority vote error correction process can be implemented, as shown in FIG. 3b, following step I of FIG. 3al frame presence detection, the

synchronization being acquired.

  Under these conditions, with reference to FIG. 3b, the above-mentioned step I can be followed by a step Jm for calculating the correlations on the word of the data Dlk and D2k,

  k denoting the index of the received frame considered.

  This calculation leads in a similar way to that of step J of FIG. 3al, it is however indicated that the values of the autocorrelation functions calculated on the data fields Dlk and D2k are noted:

ACF (Dlk) = Slk and ACF (D2k) = S2k.

  The detection process with error correction by majority vote is then carried out taking account of the index k of the frame considered with respect to the value M

  defining the depth of the majority vote introduced.

  For this purpose, the step Jvm is followed by a step Kvm for comparing the inferiority of the value of frame index k to the value M. For a succession of frames of successive index k, the resulting values Sx, Sy of the autocorrelation functions of each calculated data field, vis-à-vis each of the spread spectrum codes or sequences used, on the frames of prior index k-1 to 1 are updated. to get the resulting values

  of these autocorrelation functions.

  Thus, on a positive response to the test Kv, the index k of the current frame received having not reached the value M, the values Sx ,, and Sy of the autocorrelation functions are updated in step Mv by summing the value of the corresponding autocorrelation function relative to the current frame of rank k. The update operation is noted: Sx = Sx + Slk Sy = Sy + S2k On the contrary, on a negative response to the inferiority comparison test Kvm, the index k of the current frame having reached the value M, the values Sx andSy of the autocorrelation functions are updated in step Lv, on the one hand, by summation of the value of the corresponding autocorrelation function relating to the current frame of rank k, and, on the other hand, by subtraction of the value of the corresponding autocorrelation function calculated on the frame of rank kM. This does not

  take into account that M successive frames.

  The maximum resulting values of the absolute value of the autocorrelation functions Sx and Sy obtained in steps Mv and Lv are then subjected to a step Nvm of test of comparison of inferiority to a false detection threshold value. The aforementioned step Nv is comparable to the step K of FIG. 3aj, the threshold value of the false detection relating, as in the case of FIG. 3a1, to the data fields D1 and D2 being noted for

this reason S12.

  The comparison to the Nv test is noted Sx and Sy <S12.

  On a positive response to the Nv test, the values resulting from the autocorrelation functions Sx andSy being less than the false detection threshold value, S12, a step

  Pv delivering a false detection signal is called.

  On the contrary, on a negative response to the Nv test, a step O0 of validation of code values as the value of

  data received, noted VD1 = D1 and VD2 = D2 is called.

  It is further indicated that the resulting values Sx and Sy can be reset to zero, when there is a sudden change in the calculated values of the frame autocorrelation functions Slk or S2k, for example

  a change in digital message transmitted and received.

  Finally, it is indicated that the value of the false detection threshold S12 can be adapted during the detection process, the choice of a high value of false detection threshold, such as the value 0.9 or a higher value, allowing the start of the detection process

  to guard against possible errors.

  Experimental tests have been carried out from sound signal recordings, called test sequences, into which digital data have been inserted in accordance with the coding method which is the subject of the present invention. The test reports below relate to the decoding process implemented in the absence of correction by majority vote,

  respectively in the presence of a correction of this type.

  The references of the recordings are given in relation to the recording and environmental conditions of the corresponding sound signal. The references of the test sequences are given below: - 12_1: recording in the meeting room (empty) 50 cm away; - 12_2: recording in the meeting room (empty) at 4 m; - 13_1: recording in an apartment at 3 m without noise; - 13_2: recording in an apartment at 3 m with pre-recorded noise at the same level as the sound level of the sound signal; - 14_1: recording in an apartment at 1.9 m with noise prerecorded at the same location but at a level lower than the sound level of the useful signal; - 15_1: recording in a "Ford Galaxy" type car, in the absence of noise, windows open for a period of 12 minutes then windows closed for a period of 13 minutes; "Ford" and "Galaxy" are registered trademarks; - 15_2: recording in "Ford Galaxy" type car with loud noise, window closed for 12 min and windows open for 12 min; - 16 1: recording in "Ford Galaxy" type car with noise, louder useful noise level, windows closed for 12.30 minutes then windows

open for 12.30 min.

  Table T3 introduced below gives the decoding parameters on the test sequence 13_2 previously mentioned with a majority vote decoding of depth M for a false detection threshold S1 = S2 = FD = 0.5 and a threshold of

synchronization equal to So = 0.2.

TABLE T3

  Sequence No. of frames No. of bits No. of frames encoded inserted encoded by 30 sec

13_2 2 916 46 656 58

  Depth of Nb-frames Nb of errors Nb of frames your decoded bits received in majority M average per 30 seconds

505 2 10

550 0 11

574 0 11

593 0 11

  The depth M of the majority vote varies from 10 to and the test sequence 13_2 is an audio frequency sequence of 25 minutes of recording made

  under the conditions mentioned above.

  With reference to the aforementioned table, it is indicated that for values M of depth of the majority vote less than 20, errors may remain. These errors can for example be eliminated by increasing the false detection threshold value, this value being able to be brought to the value FD = 0.8 for example with

  however, a lower number of decoded frames.

  Thus, increasing the depth M of the majority vote makes it possible to decode a larger number of frames which were however identified as giving rise to a false detection for lower values of M. The following table, table T4 introduced below, provides a reminder of the performances obtained from the various test sequences previously mentioned.

in the description.

  The test sequences corresponding to these recordings consisted of sequences of minutes each. The depths of correction by majority vote were respectively M = 0, no correction,

  each frame being decoded individually, M = 10 and M = 50.

  During the tests carried out, the useful data detection threshold Si and S2 was equal to 0.2 while the

  false detection threshold FD = So was 0.5.

TABLE T4

  Type No. of frames No. of TBR errors (%) No. of decoded bit frames / 30s sequences M = 0 / M = 10 / M = 50 M = 0 / M = 10 / M = 50 M = 0 / M = 10 / M = 50 M = 0 / M = 10 / M = 50

  12_1 1763/2108/2112 1258/0/0 60/72/72 35/42/42

  12_2 1916/2189/2192 1128/0 / 065/75/75 38/43/43

  13_1 1422/1683/1683 493/04/049/58/58 28/33/33

  13_2 412/505/610 653/0/0 14/17/21 8/10/12

  14_1 677/844/957 781/0/0 23/29/33 13/16/19

  _1 773/956/1043 967/0/0 27/33/36 15/19/20

  -2 558/750/838 926/2/13 20/26/29 11/15/16

  16_1 850 / 1048/11601295/1/19 29/36/40 17/20/23

  In this table TBR, expressed in%, designates the good reception rate, that is to say the ratio of the number of decoded frames to the number of encoded frames. The number of frames per 30 seconds, noted Nb of frames 30 / s, indicates the

  number of frames decoded on average over 30 seconds.

  By observing table 4 above, it is indicated that the choice of a majority voting depth M = 50 compared to M = 10 makes it possible to significantly increase the number of frames decoded in the test sequences richer in codes or spreading sequences, i.e. the sequences 13_2, 14_1, 15_1 and 15_2. A priori, the choice of M = 50 is justified in the context of the aforementioned tests by the need to meet the criterion of receiving at least one code every 30 seconds. The number of errors / bit noted during decoding is significantly reduced by the correction of errors by majority vote. We note however that for a depth M = 10, two errors on the sequence 15_2 and an error on the sequence 16_1 were highlighted. It is the same for a depth M = 50, thirteen errors on the sequence 15_2 and nineteen on the sequence 16_1 having been identified. Under these conditions, and with reference to these results, it is indicated that the threshold values can be raised, the false detection threshold FD being able to be raised to 0.8 or 0.9 to keep a margin of safety. However, the increase in this threshold value has the consequence of transforming the result of the decoding of a certain number of frames decoded and accepted as valid into frames declared as subject to false detection. The number

  of decoded frames is then reduced.

  A more detailed description of a

  coding of digital information in a sound signal allowing the implementation of the coding and decoding method, object of the present invention, previously

  described in the description, will now be given in

  connection with FIG. 4a1 and the following figures.

  The device for coding digital information in an audiofrequency signal and consequently in a sound signal, object of the present invention, as shown in FIG. 4al, aims to constitute digital information in the form of digital spread spectrum frames. fixed length as described

  previously in the description and shown in figure

2a3.

  The digital spread spectrum frames consequently comprise a synchronization word MS, constituting the field Do of the abovementioned frame, and at least one useful data field, the data fields D1, D2 previously.

described.

  Referring to FIG. 4a1, it is indicated that the coding device which is the subject of the present invention comprises at least one module 1 for processing from a spread spectrum sequence of the useful data field of each digital message, each message numeric being constituted by the original message data fields dl and d2 previously mentioned in the

  description, to generate a digital spectrum message

  spread represented by each frame previously mentioned

in the description.

  The device for coding digital information in a sound signal, object of the present invention, further comprises a module 2 for inserting the digital message with spread spectrum in at least one frequency sub-band of an audiofrequency signal AS, in order to generate the marked digital audio signal, denoted AS *. Of course, the module 2 for inserting the spread spectrum digital message into a frequency sub-band of the audiofrequency signal may also include a transmission stage ensuring the radio transmission of the

  digital audio signal marked AS * above.

  More specifically, it is indicated that the module 2 for inserting the spread spectrum digital message in at least one frequency sub-band of the audio-frequency signal AS operates the aforementioned insertion from a carrier phase modulation. , the carrier CSBi of central frequency fi, and insertion of the carrier modulated by phase modulation from the resulting sequence EM into the filtered audio-frequency signal AS-, in which a frequency band of central frequency fi has been removed by filtering, as will be described

  later in the description.

  Finally, the device for coding digital information in a sound signal, object of the present invention, as shown in FIG. 4a1, further comprises a module 3 for successive reading of each frame of fixed length, that is to say say useful data of an original digital message constituted for example by useful data fields do and dl, these useful data fields having to be inserted into the sound signal in accordance with the coding method object of the present invention. The module 3 for successive reading of each frame of fixed length makes it possible to generate the digital audio signal marked AS * and of course the repetitive message inserted in the sound signal under the conditions

which will be explained below.

  As shown in FIG. 4a, it is indicated that the module 1 for processing from a spread spectrum sequence of the useful data field comprises a module for memorizing and reading at least one

spectrum spreading sequence.

  As shown in FIG. 4a1, the module for memorizing and reading the spectrum spreading sequences can comprise a first module 10 for memorizing a specific spectrum spreading sequence constituting the synchronization word, this sequence of specific spectrum spread being designated by a_d (t). In addition, the module for storing and reading spread spectrum sequences comprises a second module for storing and reading spread spectrum sequences, marked with the reference 11, and allowing the coding of useful data, from specific spectrum spread sequences, noted

code_ik (t), and code qk (t).

  As previously mentioned in the

  description, it is indicated that the spreading sequences of

  spectrum stored in module 10 preferably correspond to a pseudo-random sequence with maximum length privileged to constitute the synchronization word MS, while the spread spectrum sequences stored in module 11 are preferably

  made up of Gold or Walsh sequences-

  Hadamard. The module 1 for processing from a spectrum spreading sequence further comprises a selection module, denoted SW0, and represented symbolically by a switch between two positions, a first position I corresponding to the reading of a specific spectrum spreading sequence a_d (t) to constitute the synchronization word, and a position II corresponding to a reading of at least one spectrum spreading sequence code_ik (t) and code_qk (t) to carry out the processing by spreading spectrum of the data fields dl and d2 and thus obtaining the useful data fields with spread spectrum Dl and D2. Of course, the selection module SW0 is piloted and controlled via the module 3 for successive reading of each frame of fixed length dl, d2, in order to effect in position I the introduction of the synchronization word MS constituted by the spectrum spreading sequence a_d (t), then the spectrum spreading treatment in position II

  to obtain the aforementioned useful data fields.

  The module 1 for processing from a spectrum spreading sequence further comprises a module 12 for multiplying the sequence of the bits representative of the useful data fields d1 and d2 by one of the spectrum spreading sequences, c ' is one of the code_ik (t) or code_qk (t) sequences. The multiplication module of the bit sequence representative of the useful data field by one of the spread spectrum sequences carries the reference 12. It allows, by switching by means of a selection module SW1 from a position I to a position II, on the one hand, the introduction of the synchronization word MS, constituted by the spectrum spreading sequence a_d (t) and, on the other hand, the spectrum spreading processing of the data fields useful for generating the resulting EM sequence as shown in figures

2a1 and 2a2 for example.

  Finally, the module 2 for inserting the spread spectrum digital message, that is to say the resulting sequence EM in one of the frequency sub-bands of the audio frequency signal AS, receives the above-mentioned resulting sequence EM. It allows, by phase modulation of the carrier CBSi of central frequency fi previously mentioned, to ensure the insertion of this modulated carrier, as mentioned previously in the

  description, then the transmission of the latter with

  the audio signal marked AS * by modulation of

  channel, such as frequency modulation for example.

  As regards the phase modulation of the aforementioned carrier, it is indicated that a selection module SW2 is provided, which is controlled in synchronism between the selection module SW0 and the selection module SW1,

  to ensure BPSK type phase modulation, i.e.

  say of type MDP2, for the insertion of the synchronization word MS in position I, respectively the phase modulation of type QSPK, that is to say the coherent phase modulation MDP4 in position II, for the insertion of data fields useful for spread spectrum

D1 and D2.

  As regards the insertion of the synchronization word, it is indicated that in an advantageous embodiment, this insertion at the level of the spread spectrum processing process and of the module for multiplying the sequence of bits 12, can be carried out by via the selection module SW1 making it possible to control the aforementioned module 12 from two selection positions I and II. The selection module SW1 allows the control of the multiplication module of the sequence of bits 12 from, either constant specific data, noted Ai = 1, Bi = 0, in fact allowing the transmission by the multiplication module of the sequence of bits 12 of the single spectrum spreading sequence ad (t) when switching the selection module SW, at position I, respectively the power supply and the control of the module for multiplying the sequence of bits 12 by the intermediate bits ai and bi of the field of useful data dl and d2 stored in the storage module 3, as will be described later in

the description.

  The module 2 for inserting the digital message with spread spectrum in at least one frequency sub-band then allows the transmission of the digital audio signal marked AS *, that is to say of the useful data fields Do, Dl and D2 constituting of each length frame

  constant previously described in the description.

  A more detailed description of module 12 of

  multiplication of the bit sequence and of the module 2 of insertion by carrier modulation will now be

  given in connection with FIG. 4a2.

  The aforementioned embodiment is remarkable in that it ensures a simple transition from the BPSK type modulation process, for the insertion of the synchronization word MS, to the QPSK type modulation process for the spread processing of spectrum of useful data by simple switching, in particular at the level of the control of the module 12 for multiplying the

  bit sequence, as will be described below.

  For this purpose, the assembly constituted by the module 12 for multiplying the sequence of the bits and the module 2 for insertion by carrier modulation may advantageously comprise a phase processing channel, denoted channel I, and a processing channel in phase quadrature, noted channel Q, ensuring the transition from one to

  the other of the modulation types, and vice versa.

  As shown in FIG. 4a2 above, the module 12 for multiplying the sequence of the bits comprises a first, 121, and a second, 122, multiplier receiving the useful data bits, on the one hand, and the spreading sequence spectrum allowing spread spectrum processing to generate the

  resulting sequence EM above.

  Thus, with reference to FIGS. 4al and 4a2, the command of the module 12 for multiplying the sequence of the bits is as follows: * Selection module SW0 in position I: - The first multiplier 121 receives a constant bit value throughout the duration of the spreading sequence corresponding to the value Ai = 1, as well as the specific spectrum spreading sequence a_d (t) to deliver the same spectrum spreading sequence

  ad (t) constituting the synchronization word MS.

  - The second multiplier 122 receives, during the same duration of the spectrum spreading sequence ad (t), a bit value Bi = 0 and, moreover, a spectrum spreading sequence of substantially zero value, the second multiplier 122

delivering a value 0.

  * SW0 selection module in position II: - The first multiplier 121 receives the bit ai read from the useful data field dl or d2 as well as the spread spectrum sequence noted code_ik (t); - The second multiplier 122 receives an arbitrary bit value bi successive to the bit value ai of the useful data fields dl or d2, as well as another spread spectrum sequence, noted

code_qk (t).

  It is understood in particular that the first, 121, and the second, 122, multiplier, then deliver on the above position II the product of the bit values ai respectively bi with the corresponding spectrum spreading sequence, code_ik (t), respectively code_qk (t),

  as shown in Figure 2a1.

  In summary, as a function of the switching position I or II of the selection module SW0, the first multiplier 121 delivers the spectrum spreading sequence a_d (t) constituting the synchronization word MS and the second multiplier 122 does not deliver any signal , respectively the value processed by spread spectrum treatment of bit ai and bit bi. It is of course understood that the first multiplier 121 constitutes one of the elements of the channel in phase I and the second multiplier 122 constitutes one of the elements of the channel in quadrature of phase Q. With regard to the module 2 of insertion by carrier modulation, it is indicated that this also includes the elements necessary for the constitution of

  in-phase and quadrature phase channels.

  Under these conditions, the aforementioned module 2 comprises a first amplitude and phase multiplier, bearing the reference 21, receiving the signal delivered by the first multiplier 121 of the module 12 for multiplying the bit sequence, as well as the carrier CSBi of the form CSBi = cos2nfit The amplitude and phase multiplier 21 delivers a phase-modulated carrier component, denoted nCBSi *. In addition, an attenuator device 23 receiving the attenuation information AM corresponding to the mask value necessary in order to make the modulated carrier not audible in the audio frequency signal AS into which the latter is inserted to constitute the digital audio signal marked AS *, is provided, this attenuator module 23 receiving the carrier component modulated in phase mentioned above and delivering a carrier component modulated in phase shaped to a summing device 25. In addition, the module 2 of insertion by carrier modulation comprises a second amplitude and phase multiplier 22, receiving the signal delivered by the multiplier 122 of the bit sequence multiplication module 12, as well as a phase quadrature carrier component, denoted -sin2ifit. The amplitude and phase multiplier 22 delivers a modulated phase quadrature carrier component, denoted 7CSBi **, to an attenuator module 24 receiving the same information for mask attenuation AM. The attenuator module 24 delivers the carrier component in modulated phase quadrature and shaped to the summing circuit 25, which restores the modulated carrier nCSBi, representative of the digital message and of the digital spread spectrum frame constituted by the synchronization word MS formed. speak

  Do field then by the useful data fields D1 and D2.

  Of course, the carrier components in cos2nfit phase respectively in phase quadrature -sin2Kfit are delivered from the same carrier generator 20 whose central frequency is adjusted from the value of central frequency fi and from a phase shifter of n / 2. An equivalent diagram of the coding device represented in FIG. 4a2 is represented in FIG. 4a3 when the successive selection modules SW0, SW1 and SW2 are in position I for the insertion of the word

synchronization in each frame.

  With reference to FIG. 4a3, it is understood in particular that the channel in phase I comprising the first multiplier 121, which in fact delivers in this switching position the spreading sequence ad (t) to the first amplitude multiplier and phase 21 , this first amplitude and phase multiplier 21 and the first shaping attenuator 23 in fact constitute the aforementioned phase I channel, the latter delivering to the summing circuit 25 the synchronization word constituted by the

  spectrum spreading sequence ad (t).

  In addition, the phase quadrature channel Q. formed by the second multiplier 122, the second amplitude and phase multiplier 22 and the second attenuator circuit 24 on the contrary does not deliver any signal to the summing circuit 25, the second attenuator circuit 24 can, in such conditions, introduce infinite attenuation for example. For this reason, channel Q in FIG. 4a3 is shown in dotted lines. FIG. 4a4 shows an equivalent diagram of the device represented in FIG. 4a2 when the selection modules SW0, SW1 and SW2 of FIG. 4a1 are in switching position II to ensure the

  spread spectrum processing of useful data.

  Under these conditions, the channels in phase I and in quadrature of phase Q make it possible to ensure the processing by spreading of spectrum of two successive bits ai and bi of the useful field of the data di and d2 previously mentioned. In this configuration, however, the phase I and quadrature Q phase channels each fully play their role in generating the phase-modulated carrier representative of the useful data fields D1 and D2. A proof of the operating mode of the coding device represented in FIGS. 4a1 and 4a2 will now be given in connection with FIGS. 4a3 and 4a4: A Transmission of the synchronization word MS The signal transmitted during the synchronization word is of the form: Relation 5 Do = AM.ad (t) cos 2K fit * Transmission of the data processed by spread spectrum: When transmitting the data processed by spread spectrum, just after the transmission of the synchronization word, the signal transmitted during the transmission of the aforementioned processed data, checks the relationship: Relation 6 D1, D2 = AM.ai.code_ik (t) .cos2nfit-AM.bi code_qk (t) .sin27fit

  In the above relationships, the expressions above

  after are defined as follows: codeik (t) and code_qk (t) denote the spreading sequences applied to the channel in phase, respectively in phase quadrature, these sequences being chosen for k E [O0, nl] we denote the number of spreading sequences available to ensure the implementation of the coding method object of the present invention, AM designates the value of the mask in the central frequency sub-band fi for modulation of the corresponding CSBi carrier, ai, bi designate a pair of successive bits to be transmitted at a given time by coherent MDP4 modulation, in this case bits b6 and b7 according to the coding rule defined above, a_d (t) designates the spreading sequence retained for

  constitute the synchronization word.

  With regard to the choice of spreading sequences, it is indicated that the performance of the coding and decoding device, object of the present invention, is significantly improved if two different spreading sequences or codes are chosen for the two. channels, channel in phase I and channel in quadrature of phase Q. It is therefore advisable to choose two spreading sequences having an intercorrelation function at offset 0 as small as possible, in order to limit the interference between the

two aforementioned routes.

  A more detailed description of a

  decoding of digital information introduced into a sound signal in accordance with the coding method object of the present invention and thanks to the implementation of the coding device previously described will now be given in connection with FIG. 5a1 and the following figures. In order to improve the reliability and robustness of the process of decoding digital information introduced into a sound signal which is the subject of the present invention, it is indicated that the process of detecting frames and digital messages with spread spectrum transmitted is essential, in order to allow to locate moments

  o the data are located in the sound signal.

  The purpose of the decoding method of the present invention is therefore to continuously scan the received sound signal, at the nomadic data extraction device, in order to detect the synchronization word MS constituted by the data field Do of

each frame transmitted.

  When the aforementioned synchronization word is effectively detected, the decoding method which is the subject of the present invention then makes it possible to decode the data.

  that follow the aforementioned synchronization word.

  The digital information decoding device object of the present invention as shown in Figure 5a1 takes into account, in a particularly advantageous manner, the fact that the synchronization word, on the one hand, MS is constituted by a pure spreading sequence , while the useful data fields D1 and D2 correspond, on the other hand, to useful data fields processed by spread spectrum, in which it is possible to highlight, by correlation, one of the sequences of spread spectrum

  used at issue to provide the aforementioned processing.

  Under these conditions, and taking into account the preceding remark, the decoding device which is the subject of the present invention is designed so that a maximum of functional elements or sub-elements used for the process of detecting the synchronization word are also used to decode the

useful data by correlation.

  Thus, and according to a particularly remarkable aspect of the decoding device which is the subject of the present invention, it comprises, as shown in FIG. 5a1, a module for detecting the synchronization word making it possible to discriminate the existence of the reception of a digital spread spectrum frame, and a module for decoding the useful data of the fields of

  useful data associated with the aforementioned synchronization word.

  In particular, it is understood that to ensure the detection of the synchronization word, the decoding device object of the present invention as shown in FIG. 5a, comprises, in particular, a correlation channel making it possible to calculate the autocorrelation function of the sequence spreading ad (t) constituting the synchronization word used on transmission to detect the latter, whereas on the contrary, the decoding device object of the present invention comprises a plurality of correlation channels constituting a bank of correlators for decoding useful data transmitted and received, of course due to the M-ary coding introduced

on the show.

  It is understood in particular that the processing circuits upstream of the correlation channel, respectively of the plurality of correlation channels constituting the correlator bank, may be common due to the similar nature of the synchronization word and of the spread spectrum data. in the frequency domain. It is also understood that from the correlation channel implemented for the decoding device object of the present invention, respectively from the plurality of correlation channels constituting a bank of correlators for the implementation of this same decoding device for the decoding of useful data with spread spectrum, this correlation channel and this bank of correlators implement modules for calculating the autocorrelation function between the digital message demodulated on this field and the spectrum spread sequence used in coding, as well as a module for comparing the value of the above-mentioned autocorrelation function with a threshold value, both for detecting the synchronization word and for decoding the data

useful for spread spectrum.

  In the case of decoding of useful data with spread spectrum, the aforementioned useful data field is assigned the decoded value of the code of this field if the value of the autocorrelation function is greater than the aforementioned threshold value, whereas a false signal

  detection is generated otherwise.

  A particularly advantageous specific embodiment comprising the common functional elements upstream of the correlation channel and of the correlation channels, constituting a bank of correlators, will now be described in conjunction with FIG. 5a1 in the case of use in emission of spectrum spreading sequences for the channel in phase, respectively in phase quadrature, a similar structure can then advantageously be provided for carrying out the

  decoding device object of the present invention.

  Under these conditions, the module for detecting the synchronization word and the module for decoding useful data can comprise, to constitute a channel in phase I, respectively in quadrature of phase Q, as represented in FIG. 5aj, a module 4 for demodulation. in baseband of the modulated and coded received carrier wave, denoted r (t), the aforementioned carrier wave corresponding of course to a carrier wave whose central frequency is the frequency fi in which the data have

been introduced to the program.

  As shown in the above-mentioned figure, the baseband demodulation module 4 can advantageously include, for each channel in phase I, respectively in quadrature of phase Q, a frequency change device 41, respectively 43, receiving, on the one hand, the received modulated carrier wave nCSBi representative of the digital spread spectrum frame and designated in FIG. 5al by r (t). Each frequency change device 41, 43 respectively receives a reference carrier wave, of the same central frequency fi but of arbitrary phase 9 with respect to the aforementioned received modulated carrier wave. The frequency change device 43 of course receives the reference carrier wave phase shifted by n / 2 and of the form -sin (2nfit + p). Finally, and in a conventional manner, the module 4 for demodulation in baseband of the received modulated and coded carrier wave comprises a low-pass filter 42, respectively 44, for the corresponding channel in phase I respectively Q receiving the signal delivered by the frequency changing device 41, respectively 43, and each delivering a signal component in baseband in phase, in phase quadrature respectively, denoted Xr and Y i

respectively.

  In a particularly advantageous manner, it is indicated that the decoding device, object of the present invention, can also include a phase-locked loop, bearing the reference 5 in FIG. 5a1, making it possible in fact to correct the phase reference of the reference carrier wave, that is to say the value p assigned to it, with respect to the modulated and coded carrier wave received. In a nonlimiting embodiment, the phase locked loop 5 can comprise a calculation unit 50 receiving the component in phase Xr, respectively in phase quadrature Yi delivered by the demodulation module 4 in

  baseband of the received modulated and coded carrier wave.

  The calculation module 50 delivers, from the aforementioned components, the value of the real components Re, respectively imaginary Im of the signal components in baseband in phase, respectively in phase quadrature. A circuit 51 then makes it possible to calculate the value of p in a conventional manner and to control a local voltage-controlled oscillator delivering to the calculation module 50 components sin (p) and cos (p) making it possible to correct the phase value p. In particular, when receiving the synchronization word MS in BPSK modulation, that is to say from the Do field, the value of p is given by the relation: Relation 7 ReIm = (Xr) 2+ (yi) 2 On the contrary, in the case of reception of useful data with spread spectrum, and therefore of a QPSK type modulation, the value of 9 is given by the relation: Relation 8 -4ReIm (Re2 -Im2) (X2_r + Y 2 - if In addition, the phase-locked loop 5 is associated with a module 6 for sub-sampling each signal component in baseband in phase and in phase quadrature or, more particularly, signal components in baseband in phase and in phase quadrature whose phase value has been corrected by the phase locked loop 5, that is to say Re and Im signals. The sub-sampling module 6 can

  be placed upstream of the phase locked loop.

  The sub-sampling module 6 can thus include

  as shown in Figure 5al, sub-

  classic type samplers, represented by a controlled switch 61 and 62 on the phase path, in phase quadrature respectively. These elements will not be described in more detail because they correspond to elements known from the state of the art. It is simply indicated that the sub-sampling ratio can correspond to the sampling of four points over the duration Tc corresponding to the chip period represented in

Figure 2a1.

  Finally, the decoding device which is the subject of the present invention as shown in FIG. 5a, comprises a module 7 for calculating the abovementioned autocorrelation functions, receiving the signal component in baseband in phase, respectively in quadrature of phase sub- sampled delivered by the sub-sampling module 6. s With regard to the module 7 for calculating the autocorrelation functions, it is indicated that this comprises, as shown in FIG. 5al, a plurality of correlation channels constituted in a bank correlators. Thus, n correlation channels are assigned to the phase I channel, respectively to the phase Q quadrature channel. The correlation channels are formed by a multiplier circuit 70i1, 70ij, 70in, further receiving a spreading sequence d transmission, each followed by a summing device, bearing the reference 71il, 71ij to 71in for the phase I channel, the correlation channels for the Q phase quadrature channel being formed in the same way from 70ql multipliers, 70qd to 70qn, and summers 71ql, 71qj to 71qn for the phase quadrature channel Q. A decision module 72 receives the signals delivered by each summator for the phase and phase quadrature channels, the decision module 72 delivering worth of

  decision under the conditions which will be explained below

  after. As shown in FIG. 5aj, one of the correlation channels, the channel 1 formed by the multiplier 70il and the summator 71il for the phase path, respectively by the multiplier 70q1 and the summator 71q, for the quadrature channel of phase, is assigned to the detection of the synchronization word MS, that is to say to the field Do, by means of a selection module for the channel in phase SW3i, respectively for the channel in quadrature of phase SW3q . These selection modules make it possible to isolate the correlation channels of rank greater than 1 by switching from position II, in which all the correlation channels are assigned to the decoding of useful data, at position I o only the aforementioned rank 1 correlation for the phase and phase quadrature channel is assigned to the

  synchronization word detection.

  Of course, the selection modules SW3i and SW3q are controlled in synchronism with the detection of the synchronization word MS for each successive frame, as will be described later in the

description.

  As regards the decision module 72, this can then be configured as a function of the switching position of the selection modules SW3i and SW3q taking into account the information delivered by the signal of the summers 71il and 71q1 alone in operation for the detection of the synchronization word, respectively of all the summers 71il to 71in, respectively 71q, to 71qn during the decoding of the useful data, as will be explained in more detail below in the

description.

  In any event, the decision module 72 delivers information relating to the existence of the detection of the synchronization word MS, that is to say of the recognition of the data field Do during the switching of the communication modules. selection SW3i and SW3q in position I, this information can be limited if necessary to a binary variable representative of

  the existence of the detection of the synchronization word MS.

  On the contrary, in switching position II, the decision module 72 delivers an estimate âj, bi of the successive input bits ai, bi after M-ary decoding carried out via the decision module 72, as well as

will be described below.

  It will be understood in particular that the decoding device which is the subject of the present invention, as represented in FIG. 5al, overcomes the major difficulty according to which the synchronization word MS only appears once per frame and that it is therefore essential to implement a particularly reliable decoding device, making it possible to minimize false detections, in order to recover

  all frames transmitted and received.

  A more detailed description of the procedure

  of the detection of the synchronization word, respectively of the decoding of the useful data, that is to say of the useful data fields D1 and D2, will now be given in connection with FIGS. 5a2 to 5a8 as regards the detection of the word synchronization, respectively FIG. 5b with regard to the decoding of the data

useful above.

  FIG. 5a2 represents, by way of illustration, an equivalent diagram of the decoding device which is the subject of the present invention, in the case where the selection modules SW3i and SW3q are in switching position I and the assembly is thus configured so as to ensure

  detection of the synchronization word MS.

  In such a configuration, the decoding device which is the subject of the present invention is permanently active, until the detection of a correlation peak, that is to say for a very large value of the autocorrelation function up to 'upon detection

  complete with the synchronization word MS.

  As soon as the synchronization word has been detected, a switching from position I to position II is carried out for the aforementioned selection modules SW3i and SW3q. This switching makes it possible to switch to the decoding configuration of the useful data according to the equivalent diagram of the decoding device object of the invention according to FIG. 5a, as represented in FIG. 5b. As soon as the step of decoding the useful data is carried out and completed, the system returns

  systematically in switching position I, i.e.

* say in phase of detection of the synchronization word MS, in order to search for the presence of the next successive frame and thus ensure the decoding process in accordance with the decoding method which is the subject of the present invention. Referring to FIG. 5a2, the decoding device object of the invention according to FIG. 5a1 has been shown, in which however the phase locking loop 5 has been omitted, it being assumed that the phase correction introduced by the loop at The aforementioned phase lock was directly introduced at the local oscillator 52, the phase value 9 thus being, in fact, compensated by a value -p, the reference carrier wave supplied to the frequency change circuits 41 and 43 thus being deemed to be in phase coherence with the modulated carrier wave

  and coded received, designated by r (t).

  After transposition into baseband, i.e. frequency change and low-pass filtering, the transposition process is carried out for each insertion sub-band, as defined in table T2, to verify that frames have or have not been inserted, one obtains, taking into account the relation 5 of the input signal r (t), expressed however in the form of a digital signal, the components of the channel in phase respectively in phase quadrature, verifying the relation: Relation 9 AM r'i (u) =. ad (u) .cos (q) AM r'q (u) =. ad (u) .sin (9) 2- In this relation, u denotes the sample rank

  current after subsampling.

  Multiplication of the value of each component in phase r'i (u), respectively in phase quadrature r'q (u), by the value of the spread spectrum sequence used in the transmission a_d (q ') for the phase path, respectively -ad (u ') for the phase quadrature channel via the multiplier 70il, respectively 70q1 then the summation by the summers 71il respectively 71ql of the values of products delivered by the multipliers 70il and 70q1l , allows to deliver for the channel in phase respectively in phase quadrature the values of the autocorrelation coefficients of the spreading sequence a_d (u) verifying the relation: Relation 10 AM Nc-1 c = 2 'ad (u). ad (u '+ u). cos (p) u = 0 AM Nc-1 s = A 'a_d (u). (- 1) .a_d (u' + u) .sin (p) u = O In the previous relation, we indicate that the summation performed by the summers 71il and 71ql is performed on the number Nc of samples per chip duration of each spread spectrum sequence that multiplies the length of the synchronization word MS. In addition, u and u 'represent the separate samples for the digital signals r'i (u) and r'q (u) with respect to the spreading sequence constituting the synchronization word ad (u'). The autocorrelation peak between the synchronization word received, represented by the components in phase r'i (u) and in quadrature of phase r'q (u), and the spreading sequence used at the transmission ad (u ') to constitute the synchronization word chosen specifically from the plurality of spreading sequences stored in the module 10 of FIG. 4aj, is obtained for u' = 0. The synchronization word detection process thus searches for the precise moment when the spreading sequence received is the most correlated with the

original code ad (u ').

  Under these conditions, as represented in FIG. 5a2, the decision module 72 can comprise a module for calculating the square of the aforementioned values c and s, bearing the references 721 to 732, a summing module 723 receiving the aforementioned calculated square values and delivering a value V_seuil which is compared with the value

  threshold So previously mentioned in the description

  by a comparator 724. The comparator 724 then delivers the binary value denoted DD0, representative of the existence of a field Do and therefore of the synchronization word corresponding to the spreading sequence used on transmission. For n'É0, we consider that at the output of each summator 71il and 71ql, the values c and s are very low. If, on the contrary, n '= 0, then a very pronounced correlation peak is obtained and the values of c and s are written according to the relation: Relation 11 AM N- N-i c = AVY a d (u). ad (u) .cos (q) a = 0 AM Nc-1 s = A. Xa d (u). (- 1) .a_d (u) .sin (p) a = O Given the calculation of the square of these values by modules 721 and 722, the summing module 723 delivers the comparison value to the threshold value So, ie the value Vseuil verifying the relation: Relation 12

V threshold = AM2.Nc2.

  The aforementioned procedure relating to detection

  of the synchronization word MS is satisfactory.

  However, and in accordance with a particularly advantageous aspect of the decoding method and device which are the subject of the present invention, it is appropriate to define a criterion which makes it possible to decide whether the value V_seuil corresponds to a correlation peak or not, with a probability of weakest possible false detection. As shown in FIG. 5a3, it can be seen that a representation of the threshold value V as a function of time shows that the amplitudes of the correlation peaks corresponding to the successive values Vseuil can be located in a very large scale of amplitude of values. This situation results from the fact that the inserted digital signal, that is to say the carrier modulated in phase nCSBi is amplitude modulated by the level of the mask AM and that the latter evolves in a

large amplitude scale.

  In FIG. 5a4, a recording of the value Vseuil on a dilated scale in amplitude has been shown, that is to say according to a so-called zoomed representation. In this representation, the relative maximum 1 in fact constitutes a correlation peak, which can be

  highlighted as follows.

  According to a specific operating mode, the value of the autocorrelation function V_seuil can advantageously be normalized with respect to the effective energy of the

signal received.

  Such normalization makes it possible to obtain a value of the autocorrelation function of between -1

and 1.

  If this value is close to -1, then the signals are anticorrelated, if it is close to the value 0, the signals are not correlated, and if it is close

  of the value 1, the signals are then correlated.

  Taking into account the abovementioned normalization process, the normalized threshold value, denoted V threshold n is written according to the relationship: Relation 13 V threshold V threshold n = - s - (Nc-1 No-1 L Er '(u) + Er '(u) (2. (4xLong_ code_début)) u = O u = O In this relation, the denominator represents the energy of the received signal and Nc = 4Long_code_début represents 4 times the

  length of bit duration, due to sub-

  4 samples per chip sampling.

  The normalized threshold value V threshold n corresponding to the threshold value of FIG. 5a3 is

shown in Figure 5a5.

  A zoom applied to FIG. 5a5 represented in FIG. 5a6 shows that a maximum clearly becomes observable in position (1) due to the normalization introduced. In accordance with the process of detecting the synchronization word allowing decoding according to the method which is the subject of the present invention, it is then appropriate to decide that all the values greater than the value

  So = 0.15 are then part of a correlation peak.

  Finally, the detection of the maximum value of the correlation peak must be carried out in order to detect the existence of the synchronization word MS with the best probability of detection and with a minimum calculation time. Due to the sub-sampling at the rate of 4 subsampled values per chip periods, that is to say by Tc chip durations, there are 7 autocorrelation values which are capable of giving a normalized threshold value greater than the value of SO threshold. In FIG. 5a7, a time diagram of the reception of the received signal r (t) is shown with the actual position of the successive chips. The positions of spreading sequences or codes shifted by the value Tc / 4 are also represented successively in FIG. 5a7 in order to calculate the normalized threshold value Vseuiln.

  The corresponding normalized threshold value obtained is represented in FIG. 5a8 for the successive shifts 0, 1,2, 3, 4 and 5 represented in FIG. 5a7. The value of the correlation peak corresponding to offset 5, lower than the previous value corresponding to offset 4, then makes it possible to trigger the switching of the selection modules SW3i and SW3q in position II to initiate the step of decoding the useful data, the value

  maximum of the correlation peak having been obtained.

  The decoding device which is the subject of the present invention will now be described during the configuration of the latter during the decoding of the useful data, following the switching of the aforementioned selection modules.

  in position II, in connection with FIG. 5b.

  In the above-mentioned figure, in the same way as in the case of FIG. 5a2 and for the same reasons, the phase-locked loop 5 has not been shown, the phase correction process being introduced in a similar manner taking into account of the phase value calculated within the framework of coherent MDP4 modulation

  mentioned previously in the description in connection

with Figure 5a1.

  In this configuration, the decoding device which is the subject of the invention, and in particular the module 7 for calculating the autocorrelation functions between the spreading code or sequence used to carry out the spread spectrum processing and of course the

  coding of M-ary type on transmission, is taken into account.

  Thus, each multiplier circuit of the channel in phase ij respectively in phase quadrature 70qj with i [l, n] or n denotes the number of spreading sequences or codes liable to be used on transmission, receives one spreading sequences previously noted code_ik (t) for the channel in phase I, respectively code_qk (t) for the channel in quadrature of phase Q with

k e [l, n].

  Consequently, the summing circuits of the channel in phase 71il to 71in respectively of the channel in quadrature of phase 71ql to 71qn make it possible to calculate under the same computation and sampling conditions of 4 points per chip period Tc the value of the function autocorrelation between each spreading code or sequence likely to be used on transmission and the useful data received, as mentioned above

  in the description. The calculation is done in a way

  similar to that performed for the detection of the synchronization word with an offset 0 with, however, two sets of distinct spreading sequences applied for the correlation channels of the channel in phase, respectively in phase quadrature. Under these conditions, the decision module 72 can advantageously comprise for each correlation channel a module for calculating a standardized threshold function Vseuiln, each channel for correlation by the corresponding calculation modules 725il to 725in and 725ql to 725qn delivering a standardized threshold value corresponding to the spreading sequence capable of being used on transmission with respect to the received signal. A module 726 receives all of the above-mentioned normalized threshold values, denoted V_seuilil à V_seuil i n respectively Vseuil q_1 à V_seuil_q_n. The module 726 then makes it possible to select the output of the correlation channel which gives the greatest value in absolute value for example and therefore the spread spectrum sequence which codes the bits b0, b2, b4 or bl, b3, b5, according to the coding rule previously described in Table T1; if this is positive, then we say that the bit b6 or b7 received is 1, otherwise, if it is negative, the bit received has

the value 0.

  On the contrary, in the case of implementing the majority vote error correction process for data detection, the estimation of the bits is carried out after summing the M above-mentioned normalized threshold values V_seuili_1 to Vseuilin respectively V_seuil_q_l to V_seuil_q_n, M designating the depth of the majority vote. The maximum value in absolute value of the summed correlations is used to designate the code or the spectrum spreading sequence which codes the three bits b0, b2, b4 or bl, b3, b5 and the sign of this value makes it possible to determine the value. bits b6 or b7. The corresponding estimated values âj (u) and bj (u) are delivered by the module 726 corresponding to the estimated values of the successive input bits b0, b2, b4, b6

bl, b3, b5, b7 respectively.

  The decoding object of the present invention, in the configuration shown in Figure 5b, is satisfactory. However, it assumes that the modulated carrier frequency nCBSi extracted from the received signal does not vary too much during the decoding of the data. However, the phase-locked loop makes it possible, at least partially, to guard against the aforementioned problem to the extent that, in the event of variation of the aforementioned frequency, the phase correction makes it possible to at least partially make up for the difference frequency mentioned. Indeed, in the case where the difference between the phase of the received signal and that of the local oscillator 50 increases, the phase correction introduced makes it possible to catch the phase value, and therefore of frequency, until the channels are reversed in phase I and in quadrature of phase Q.

Claims (19)

RESELL I CATIONS   1. Method for encoding digital information in a sound signal, generated by the loudspeaker of a target receiver, excited by an audio frequency signal, and received by a reception microphone, by means of a reverberant environment, characterized in that it consists in: - establishing said digital information in the form of frames of fixed length, each frame comprising at least one useful data field, said frames constituting a digital message being subjected to processing from a sequence spread spectrum over a determined number of bits to generate a spread spectrum digital message; - inserting said digital message into at least one frequency sub-band of said audio frequency signal, by phase modulation around a carrier inserted into this frequency sub-band, from said digital spread spectrum message; - repeating the insertion of said spread spectrum digital message for at least one frame to generate a message   repetitive digital inserted in said sound signal.   2. Method according to claim 1, characterized in that said step of inserting said digital message consists of inserting said frames in at least one of two low frequency sub-bands whose central frequency is between 2.5 kHz and 4 kHz and in at least one of four high frequency sub-bands, between 4 kHz and the cut-off frequency of said sound signal.   3. Method according to claim 1 or 2, characterized in that each frame comprises at least: - a synchronization word; - a field of useful data.   4. Method according to one of claims 1 to 3,   characterized in that said step consisting in establishing said digital information consists in subjecting each frame to an M-ary coding to generate a series of   symbols obtained by coding p bits per symbol.   5. Method according to claims 3 and 4,   characterized in that said spread spectrum processing step comprises at least applying to said auxiliary sound signal direct sequence spread spectrum sequences, the type of spread sequence being chosen as a function of the field of each   constitutive frame of said digital signal.   6. Method according to claim 5, characterized in that said spectrum spreading sequences are chosen from the group of pseudo-random sequences,   Gold sequences, Walsh-Hadamard sequences.   7. Method for decoding a digital message with spread spectrum coded and inserted in a sound signal, transmitted and restored by the acoustic channel of a target receiver, according to the coding method according to claim 1, characterized in that, on reception of at least one spread spectrum digital message consisting of a digital frame, each digital frame being subjected to demodulation and decoding from a spread spectrum sequence, said method consists in: - calculating for at at least one of the data fields of each digital frame the value of the autocorrelation function between the demodulated message on this field and the spreading sequence used on transmission; - comparing said value of the autocorrelation function with a determined threshold value; - retain as the code value of the data field the value of this field if the value of the autocorrelation function is greater than said determined threshold value;   - to generate a false detection signal otherwise.   8. A decoding method according to claim 7, characterized in that for successive frames containing a specific synchronization word, said determined threshold value, constituting a synchronization threshold value, is between 0.20 and 0.26,   depending on the nature of the acoustic channel.   9. A decoding method according to claim 7, characterized in that for successive frames comprising at least one useful data field, said determined threshold value, constituting a false detection threshold value, is between 0.03 and 0 , 9 according to the sequences transmitted by the acoustic channel and the application of a process of   error correction by majority vote.   10. Method according to one of claims 7 to 9,   characterized in that said decoding method further includes an error correction process of the type to majority vote.   11. A decoding method according to claim 10, characterized in that for said majority vote type error correction process, said determined threshold value relating to said specific synchronization word is taken equal to the lower value of 0.20 the range of values, so as to allow the decoding of a larger significant number of frames, and in that the decoding of the useful data is carried out by correlation of N = 2P codes with said useful data, the signed levels of the peaks of correlation of each code and the useful data being integrated on a number M of words, the number M of which defines the depth of the vote majority.   12. Device for coding digital information in a sound signal, generated by the loudspeaker of a target receiver, excited by a marked audio-frequency signal, this sound signal being received by a microphone via a reverberant environment , this digital information being established in the form of digital frames of fixed length, each frame comprising a synchronization word and at least one useful data field, characterized in that it comprises at least: - means for processing from a spectrum spreading sequence of the useful data field of each digital message, for generating a spread spectrum digital message; - means for inserting said spread spectrum digital message into at least one frequency sub-band of an audio-frequency signal, to generate said marked digital audio signal; means for successively reading each frame of fixed length, which makes it possible to repeat the insertion of said digital message with spread spectrum to generate said marked digital audio signal and said message   repetitive inserted in said sound signal.   13. Device according to claim 12, characterized in that said processing means, from a spread spectrum sequence of the useful data field comprise at least: - storage and reading means of at least one spectrum spreading sequence among a plurality of spectrum spreading sequences; means for multiplying the sequence of bits representative of said useful data field by one of the spread spectrum sequences, which allows   to generate a spread spectrum digital message.   14. Device according to one of claims 12   or 13, characterized in that said means for inserting said digital spread spectrum message into at least one frequency sub-band of an audiofrequency signal comprise at least one channel for inserting said digital spread spectrum message comprising: means generating at least one carrier wave, the center frequency of which is equal to the center frequency of one of the frequency sub-bands of the digital audio signal; - phase modulating means receiving, on the one hand, said carrier wave, and on the other hand, said digital message with spread spectrum and delivering a modulated carrier wave; - shaping means making it possible to adapt the amplitude level of the modulated carrier wave, to generate a calibrated modulated carrier wave intended to be inserted in said frequency sub-band of the audio signal.   15. Device according to claim 14, characterized in that it comprises: - a first channel for inserting said spread spectrum digital message; - a second switched insertion channel of said spread spectrum digital message; - switching means of said second switched insertion channel, said switching means making it possible to deliver: * in a first switching position, a pure spectrum spreading sequence to said phase modulator means of said first insertion channel and a carrier wave and a digital message with spectrum spread at the substantially zero value to said phase modulator means of said second switched insertion channel, for generating by BPSK modulation of said carrier wave said synchronization word; * in a second switching position, a reference carrier wave and a first digital spectrum message spread from a first spectrum spreading sequence to said phase modulator means of the first insertion channel, respectively, a carrier wave, of the same frequency as the reference carrier wave but in phase quadrature therewith and a second digital spectrum message spread from a second spread spectrum sequence to said modulating means of said second switched insertion path , to generate by QPSK modulation of said reference carrier wave said data word, constituting the   spread spectrum digital message.   16. Device for decoding digital information introduced into a sound signal, in the form of a digital message with coded spread spectrum and inserted into this sound signal by insertion of a modulated and coded carrier, this sound signal being transmitted and reproduced by the reverberant acoustic channel of a target receiver, this spread spectrum digital message consisting of a fixed length digital frame comprising a synchronization word consisting of a pure spectrum spread sequence and at least one useful data field coded at starting from a spectrum spreading sequence, these sequences being chosen from a plurality of spectrum spreading sequences, characterized in that said decoding device comprises: - means for detecting said synchronization word making it possible to discriminate between existence of reception of a digital frame; and means for decoding useful data from said useful data field associated with said synchronization word, said decoding means comprising at least: means for calculating the autocorrelation function between the digital message demodulated on this field and the sequence spread spectrum used for coding; ò means for comparing the value of said autocorrelation function with a threshold value, said useful data field being assigned the decoded value of the code of this field if the value of said autocorrelation function is greater than said value of threshold, a false detection signal being spawned otherwise.   17. Device according to claim 16, characterized in that said means for detecting said synchronization word and said means for decoding useful data comprise, constituting a channel in phase respectively in phase quadrature: - baseband demodulation means of the modulated and coded carrier wave received, respectively of a modulated and coded carrier wave received in phase quadrature, with respect to a reference carrier wave each delivering a signal component in baseband in phase, respectively in phase quadrature; means for sub-sampling each signal component in phase band in phase, respectively in phase quadrature, delivering a signal component in base band in phase, respectively in quadrature of sub-sampled phase.   18. Device according to claim 17, characterized in that said means for detecting said synchronization word also comprise: first and second decoding means receiving said signal component in baseband in phase and said spectrum spreading sequence pure, constituting the synchronization word, respectively said phase quadrature baseband signal component and said pure spectrum spreading sequence constituting the synchronization word, the first and second decoding means delivering a phase decoded signal, respectively in phase quadrature; - first and second summing means receiving the phase decoded signal, respectively in phase quadrature and delivering the components in phase, respectively in phase quadrature, of the value of the autocorrelation function of the pure spectrum spreading sequence with the synchronization word of the received frame; - first and second means for calculating the square of the components in phase, respectively in phase quadrature of the autocorrelation function; - summing means receiving the value of the square of the components in phase, respectively in phase quadrature of the autocorrelation function and delivering the value of said autocorrelation function to be compared at said threshold value.   19. Device according to Claim 17, characterized in that the said means for decoding the useful data comprise, for each channel in phase, respectively in phase quadrature, a plurality of n correlation channels, the number of the said plurality of sequences is designated. spread spectrum among which the spectrum spread sequence used for transmission is retained, each correlation channel comprising: - a decoding module receiving said signal component in baseband in phase or in phase quadrature and one of the n spread spectrum sequences of the plurality of n spread spectrum sequences and outputting a decoded baseband signal for said one considered spread spectrum sequence; - summing means over the duration of said spreading sequence of said decoded baseband signal receiving said decoded baseband signal and delivering a value of the autocorrelation function for said a spectrum spreading sequence considered; means for calculating error correction of the majority vote type, on M useful data fields and successive frames, M designating the depth of the majority vote, making it possible to determine the maximum value of the value of the autocorrelation function for each spread spectrum sequence likely to be used and the value of the useful data field considered.